No matter what people say, doing ecology is hard! Natural systems are, after all, complex. There are lots of factors that affect ecological processes, and teasing them out can be a challenge. Hence, trying to decide what to study can be a daunting task for any new grad student. Funnily enough, however, modern ecology is not data (or technique or computation) limited – it’s often ideas limited. If you are lucky (like I was), you get to study the "low-hanging fruit" - the questions that need to be answered first before you can move on to deeper understanding. Do grassland plants have soil stored seeds? Is germination stimulated by fire? How big does a gap in the grassland canopy need to be for recruitment of forbs? Once these questions have been picked off, you'll have to think more critically about the next round of obvious things to ask. And this is the challenge.
Hence, when choosing a research project, it's advisable to first read widely, see what’s going on in your field (and related fields), and identify some general question or idea that you can address in your own system? This means being inquisitive and wanting to know where the field is at and how you might contribute meaningfully.
So, how important should this question be?
That is a difficult question to answer. Publishing metrics (Impact Factors, H-scores, etc) and funding trends mean that you should have your eye "on the game" if you want to make ecological research your career. But this doesn't guarantee you'll do good science. Good science involves carefully thought out hypotheses, and well-designed tests of these hypotheses. And to succeed, I think it is critical that you passionately believe in the research you are doing - you own it! Hence, jumping on the latest "hot topic" in ecology (think the productivity-richness debate) might make your research seem current, but it is just as likely to move quickly on. Good ecological science, however, will stand the test of time.
There are two general observations I'd make when it comes to developing research questions:
First, avoid testing poorly developed hypotheses / theories with data you really don't understand and which wasn’t collected for the purpose. Trying to make your data fit a question isn't good science. That said, the recent surge in meta-analyses shows that when such data do exist, they can provide powerful insights that site by site studies just can't achieve.
Second, selecting a topic based on the choice of tool, not because of the question, is a no-no in my opinion. I’ve seen papers that seem to imply their study is important simply because they used [insert latest fad or bandwagon here]. Bandwagons that depend heavily on technical proficiency might make your research look great, but from my perspective, it is much easier to learn a new skill than to do original, ground-breaking science.
Of course, this is my personal view on the matter and you may disagree. There is no one or right approach here. That said, I'm a question-driven person. Tools are just that: tools! Good luck navigating a research thesis - it really is one the best things you'll ever do - particularly if you find a topic that grabs you and acknowledges those who came before you!!
I'll leave you with a recent comment I saw from Steve Packard (in Ian Lunt's excellent blog) - I thought it was apt (he was talking about the early days of restoration ecology):
"restoration was in the early Wright Brothers stage of flight. We were barely getting off the ground, but we were tackling the most fun and fundamental questions...............in the Wright Brothers’ day, learned academics were also trying to explore flight, but the Wrights made the key discoveries because they had the hammers and wrenches in their own hands and did their own flying."
